Commit Graph

116 Commits

Author SHA1 Message Date
Evan Cheng
0cc83b6e85 Forgot this.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@47510 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-23 00:46:11 +00:00
Evan Cheng
d70dbb5d62 Enable re-materialization of instructions which have virtual register operands if
the definition of the operand also reaches its uses.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@47475 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-22 09:24:50 +00:00
Roman Levenstein
8dd25288f9 New helper function getMBBFromIndex() that given an index in any instruction of an MBB returns a pointer the MBB. Reviewed by Evan.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@47267 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-18 09:35:30 +00:00
Evan Cheng
c8d044e4f7 - Removing the infamous r2rMap_ and rep() method. Now the coalescer will update
register defs and uses after each successful coalescing.
- Also removed a number of hacks and fixed some subtle kill information bugs.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@47167 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-15 18:24:29 +00:00
Evan Cheng
b1f6f91393 Simplify.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@47059 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-13 09:18:16 +00:00
Evan Cheng
70071434ae Initial support for copy elimination by commuting its definition MI.
PR1877.
A3 = op A2 B0<kill>                                                                                                                                                                            
...                                                                                                                                                                                     
B1 = A3      <- this copy                                                                                                                                                                      
...                                                                                                                                                                                          
   = op A3   <- more uses                                                                                                                                                                      
                                                                                                                                                                                                 
==>                                                                                                                                                                                             
                                                                                                                                                                                            
B2 = op B0 A2<kill>                                                                                                                                                                            
...                                                                                                                                                                                          
B1 = B2      <- now an identify copy                                                                                                                                                           
...                                                                                                                                                                                          
   = op B2   <- more uses

This speeds up FreeBench/neural by 29%, Olden/bh by 12%, oopack_v1p8 by 53%.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@47046 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-13 03:01:43 +00:00
Dan Gohman
6f0d024a53 Rename MRegisterInfo to TargetRegisterInfo.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46930 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-10 18:45:23 +00:00
Chris Lattner
84bc5427d6 Rename SSARegMap -> MachineRegisterInfo in keeping with the idea
that "machine" classes are used to represent the current state of
the code being compiled.  Given this expanded name, we can start 
moving other stuff into it.  For now, move the UsedPhysRegs and
LiveIn/LoveOuts vectors from MachineFunction into it.

Update all the clients to match.

This also reduces some needless #includes, such as MachineModuleInfo
from MachineFunction.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@45467 91177308-0d34-0410-b5e6-96231b3b80d8
2007-12-31 04:13:23 +00:00
Chris Lattner
7ed47a1335 Don't attribute in file headers anymore. See llvmdev for the
discussion of this change.  Boy are my fingers tired. ;-)



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@45411 91177308-0d34-0410-b5e6-96231b3b80d8
2007-12-29 19:59:42 +00:00
Evan Cheng
22f07ffd27 Switch over to MachineLoopInfo.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@44838 91177308-0d34-0410-b5e6-96231b3b80d8
2007-12-11 02:09:15 +00:00
Evan Cheng
5ef3a04b54 Fix for PR1831: if all defs of an interval are re-materializable, then it's a preferred spill candiate.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@44644 91177308-0d34-0410-b5e6-96231b3b80d8
2007-12-06 00:01:56 +00:00
Evan Cheng
018f9b020b - Mark last use of a split interval as kill instead of letting spiller track it.
This allows an important optimization to be re-enabled.
- If all uses / defs of a split interval can be folded, give the interval a
  low spill weight so it would not be picked in case spilling is needed (avoid
  pushing other intervals in the same BB to be spilled).


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@44601 91177308-0d34-0410-b5e6-96231b3b80d8
2007-12-05 03:22:34 +00:00
Evan Cheng
d64b5c82b9 Add a argument to storeRegToStackSlot and storeRegToAddr to specify whether
the stored register is killed.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@44600 91177308-0d34-0410-b5e6-96231b3b80d8
2007-12-05 03:14:33 +00:00
Evan Cheng
aee4af68ae Remove redundant foldMemoryOperand variants and other code clean up.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@44517 91177308-0d34-0410-b5e6-96231b3b80d8
2007-12-02 08:30:39 +00:00
Evan Cheng
e62f97c094 Allow some reloads to be folded in multi-use cases. Specifically testl r, r -> cmpl [mem], 0.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@44479 91177308-0d34-0410-b5e6-96231b3b80d8
2007-12-01 02:07:52 +00:00
Evan Cheng
cddbb83ea8 Do not fold reload into an instruction with multiple uses. It issues one extra load.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@44467 91177308-0d34-0410-b5e6-96231b3b80d8
2007-11-30 21:23:43 +00:00
Evan Cheng
1953d0cb7d Fix a major performance issue with splitting. If there is a def (not def/use)
in the middle of a split basic block, create a new live interval starting at
the def. This avoid artifically extending the live interval over a number of
cycles where it is dead. e.g.

bb1:
       = vr1204   (use / kill) <= new interval starts and ends here.
...
...
vr1204 =          (new def)   <= start a new interval here.
       = vr1204   (use)


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@44436 91177308-0d34-0410-b5e6-96231b3b80d8
2007-11-29 10:12:14 +00:00
Evan Cheng
0cbb1164b3 Fixed various live interval splitting bugs / compile time issues.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@44428 91177308-0d34-0410-b5e6-96231b3b80d8
2007-11-29 01:06:25 +00:00
Evan Cheng
cada245d06 Recover compile time regression.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@44386 91177308-0d34-0410-b5e6-96231b3b80d8
2007-11-28 01:28:46 +00:00
Evan Cheng
81a0382181 Live interval splitting:
When a live interval is being spilled, rather than creating short, non-spillable
intervals for every def / use, split the interval at BB boundaries. That is, for
every BB where the live interval is defined or used, create a new interval that
covers all the defs and uses in the BB.

This is designed to eliminate one common problem: multiple reloads of the same
value in a single basic block. Note, it does *not* decrease the number of spills
since no copies are inserted so the split intervals are *connected* through
spill and reloads (or rematerialization). The newly created intervals can be
spilled again, in that case, since it does not span multiple basic blocks, it's
spilled in the usual manner. However, it can reuse the same stack slot as the
previously split interval.

This is currently controlled by -split-intervals-at-bb.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@44198 91177308-0d34-0410-b5e6-96231b3b80d8
2007-11-17 00:40:40 +00:00
Hartmut Kaiser
ffb15de60e Fixed a compilation error.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@44083 91177308-0d34-0410-b5e6-96231b3b80d8
2007-11-13 23:04:28 +00:00
Evan Cheng
f2fbca68f8 Refactor some code.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@44010 91177308-0d34-0410-b5e6-96231b3b80d8
2007-11-12 06:35:08 +00:00
Evan Cheng
c92da3882e There are times when the coalescer would not coalesce away a copy but the copy
can be eliminated by the allocator is the destination and source targets the
same register. The most common case is when the source and destination registers
are in different class. For example, on x86 mov32to32_ targets GR32_ which
contains a subset of the registers in GR32.

The allocator can do 2 things:
1. Set the preferred allocation for the destination of a copy to that of its source.
2. After allocation is done, change the allocation of a copy destination (if
   legal) so the copy can be eliminated.

This eliminates 443 extra moves from 403.gcc.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@43662 91177308-0d34-0410-b5e6-96231b3b80d8
2007-11-03 07:20:12 +00:00
Evan Cheng
cf1d7256d0 Remove unnecessary include.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@43117 91177308-0d34-0410-b5e6-96231b3b80d8
2007-10-18 07:47:55 +00:00
Evan Cheng
a5bfc97da7 Apply Chris' suggestions.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@43069 91177308-0d34-0410-b5e6-96231b3b80d8
2007-10-17 06:53:44 +00:00
Evan Cheng
4ca980e7f9 Clean up code that calculate MBB live-in's.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@43060 91177308-0d34-0410-b5e6-96231b3b80d8
2007-10-17 02:10:22 +00:00
Evan Cheng
32dfbeada7 EXTRACT_SUBREG coalescing support. The coalescer now treats EXTRACT_SUBREG like
(almost) a register copy. However, it always coalesced to the register of the
RHS (the super-register). All uses of the result of a EXTRACT_SUBREG are sub-
register uses which adds subtle complications to load folding, spiller rewrite,
etc.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@42899 91177308-0d34-0410-b5e6-96231b3b80d8
2007-10-12 08:50:34 +00:00
Evan Cheng
f3bb2e65d1 Use pool allocator for all the VNInfo's to improve memory access locality. This reduces coalescing time on siod Mac OS X PPC by 35%. Also remove the back ptr from VNInfo to LiveInterval and other tweaks.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@41729 91177308-0d34-0410-b5e6-96231b3b80d8
2007-09-05 21:46:51 +00:00
Evan Cheng
35b35c5c32 Add a variant of foldMemoryOperand to fold any load / store, not just load / store from / to stack slots.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@41597 91177308-0d34-0410-b5e6-96231b3b80d8
2007-08-30 05:52:20 +00:00
Evan Cheng
7ecb38be0a Change LiveRange so it keeps a pointer to the VNInfo rather than an index.
Changes related modules so VNInfo's are not copied. This decrease
copy coalescing time by 45% and overall compilation time by 10% on siod.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@41579 91177308-0d34-0410-b5e6-96231b3b80d8
2007-08-29 20:45:00 +00:00
Evan Cheng
549f27d307 Re-implement trivial rematerialization. This allows def MIs whose live intervals that are coalesced to be rematerialized.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@41060 91177308-0d34-0410-b5e6-96231b3b80d8
2007-08-13 23:45:17 +00:00
David Greene
2513330de8 Factor live variable analysis so it does not do register coalescing
simultaneously.  Move that pass to SimpleRegisterCoalescing.

This makes it easier to implement alternative register allocation and
coalescing strategies while maintaining reuse of the existing live
interval analysis.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@37520 91177308-0d34-0410-b5e6-96231b3b80d8
2007-06-08 17:18:56 +00:00
Evan Cheng
2f6cb2b14c Fix for PR1406:
v1 =
r2 = move v1
   = op r2<kill>
...
r2 = move v1
   = op r2<kill>

Clear the first r2 kill if v1 and r2 are joined.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@37050 91177308-0d34-0410-b5e6-96231b3b80d8
2007-05-14 21:10:05 +00:00
Nick Lewycky
ecd94c804a Fix typo in comment.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@36873 91177308-0d34-0410-b5e6-96231b3b80d8
2007-05-06 13:37:16 +00:00
Devang Patel
1997473cf7 Drop 'const'
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@36662 91177308-0d34-0410-b5e6-96231b3b80d8
2007-05-03 01:11:54 +00:00
Devang Patel
3e15bf33e0 Use 'static const char' instead of 'static const int'.
Due to darwin gcc bug, one version of darwin linker coalesces
static const int, which defauts PassID based pass identification.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@36652 91177308-0d34-0410-b5e6-96231b3b80d8
2007-05-02 21:39:20 +00:00
Devang Patel
794fd75c67 Do not use typeinfo to identify pass in pass manager.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@36632 91177308-0d34-0410-b5e6-96231b3b80d8
2007-05-01 21:15:47 +00:00
Evan Cheng
24a3cc4c83 Fix for PR1306.
- A register def / use now implicitly affects sub-register liveness but does
not affect liveness information of super-registers.
- Def of a larger register (if followed by a use later) is treated as
read/mod/write of a smaller register.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@36434 91177308-0d34-0410-b5e6-96231b3b80d8
2007-04-25 07:30:23 +00:00
Evan Cheng
faf05bbaea Don't populate TryAgainList when coalescing only physical registers with virtual registers.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@36240 91177308-0d34-0410-b5e6-96231b3b80d8
2007-04-18 02:30:19 +00:00
Evan Cheng
20b0abc24f Copy coalescing change to prevent a physical register from being pin to a
long live interval that has low usage density.
1. Change order of coalescing to join physical registers with virtual
   registers first before virtual register intervals become too long.
2. Check size and usage density to determine if it's worthwhile to join.
3. If joining is aborted, assign virtual register live interval allocation
   preference field to the physical register.
4. Register allocator should try to allocate to the preferred register
   first (if available) to create identify moves that can be eliminated.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@36218 91177308-0d34-0410-b5e6-96231b3b80d8
2007-04-17 20:32:26 +00:00
Evan Cheng
bcfd4665b5 Ugh. Copy coalescer does not update register numbers.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@35600 91177308-0d34-0410-b5e6-96231b3b80d8
2007-04-02 18:49:18 +00:00
Evan Cheng
88d1f587d2 Track all joined registers and eliminate unneeded kills after all joining are done.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34767 91177308-0d34-0410-b5e6-96231b3b80d8
2007-03-01 02:03:03 +00:00
Evan Cheng
edeffb37dc Joining an interval with a dead copy instruction. Shorten the live range to the last use.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34645 91177308-0d34-0410-b5e6-96231b3b80d8
2007-02-26 21:37:37 +00:00
Evan Cheng
7d35c0e2b9 Copy and paste silliness.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34507 91177308-0d34-0410-b5e6-96231b3b80d8
2007-02-22 23:52:23 +00:00
Evan Cheng
30cac02a92 Remove unnecessary isKill properties if a live range has been lengthened due to coalescing.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34503 91177308-0d34-0410-b5e6-96231b3b80d8
2007-02-22 23:03:39 +00:00
Jim Laskey
9b25b8ca24 Allow for live in registers for eh landing pads.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34475 91177308-0d34-0410-b5e6-96231b3b80d8
2007-02-21 22:41:17 +00:00
Evan Cheng
b371f457b0 Re-apply my liveintervalanalysis changes. Now with PR1207 fixes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34428 91177308-0d34-0410-b5e6-96231b3b80d8
2007-02-19 21:49:54 +00:00
Reid Spencer
a284cbf667 For PR1207:
Revert patches that caused the problem. Evan, please investigate and reapply
when you've discovered the problem.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34399 91177308-0d34-0410-b5e6-96231b3b80d8
2007-02-19 03:20:00 +00:00
Evan Cheng
aa7976199c - Changes how function livein's are handled, they now have a start index of 0.
- When coalescing a copy MI, if its destination is "dead", propagate the
  property to the source MI's destination if there are no intervening uses.
- Detect dead function live-in's and remove them.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34383 91177308-0d34-0410-b5e6-96231b3b80d8
2007-02-17 11:15:40 +00:00
Evan Cheng
61de82d885 Use BitVector instead of vector<bool> which can be extremely slow.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34302 91177308-0d34-0410-b5e6-96231b3b80d8
2007-02-15 05:59:24 +00:00